NL-15-029, Summary of Analysis Results Based on Corrected Format 2 Data Files Dated March 2015
| ML15252A244 | |
| Person / Time | |
|---|---|
| Site: | Indian Point  |
| Issue date: | 04/06/2015 |
| From: | AKRF |
| To: | Indian Point, License Renewal Projects Branch 2 |
| References | |
| NL-15-029 | |
| Download: ML15252A244 (18) | |
Text
Summary of Analysis Results Based on Corrected Format 2 Data Files Dated March 2015 Update of Aquatic Impact Analyses Presented in NRCs FSEIS (December 2010)
Regarding Potential Impacts of Operation of Indian Point Units 2 and 3 Prepared for:
INDIAN POINT ENERGY CENTER 450 Broadway, Suite 1 Buchanan, NY 10511 Prepared by AKRF, Inc.
7250 Parkway Drive, Suite 210 Hanover, MD 21076 April 2015
Table 3. Competing Models Used To Characterize the Standardized River Segment 4 FSS Population Trends of YOY Fish Density Using a 3-Year Moving Average (updated FSEIS Table I-9).
Species Linear Regression Segmented Regression MSE Slope Std Err p-value MSE Slope 1 Join Slope 2 of Slope Point Estimate Lower Upper Lower Upper 95% CL 95% CL 95% CL 95% CL Alewife 0.875 -0.055 0.026 0.047 0.863 -2.189 3.276 1989 -0.132 -0.010 American Shad 0.271 -0.117 0.014 0.000 0.249 -2.156 0.780 1988 -0.143 -0.078 Atlantic Tomcod 0.666 -0.082 0.023 0.001 0.410 -1.081 -0.239 1991 -0.080 0.023 Bay Anchovy 0.601 -0.088 0.021 0.000 0.527 -1.456 2.814 1989 -0.158 -0.063 Blueback Herring 0.877 -0.054 0.026 0.048 0.083 -5.262 -3.565 1988 -0.027 0.010 Bluefish 0.925 -0.046 0.027 0.099 0.587 0.080 1.506 1990 -0.160 -0.045 Hogchoker 0.434 -0.104 0.018 0.000 0.275 -0.271 -0.127 2001 -0.070 0.138 Rainbow Smelt 0.623 -0.086 0.022 0.001 0.535 -0.193 0.534 1992 -0.188 -0.060 Striped Bass 0.745 -0.073 0.024 0.006 Failed to Converge Weakfish 0.810 -0.064 0.025 0.017 Failed to Converge White Catfish 0.941 -0.042 0.027 0.128 0.892 -0.144 0.363 1995 -0.206 0.007 White Perch 0.838 -0.060 0.025 0.026 0.656 -0.543 -0.023 1995 -0.062 0.104 Table 4. River Segment 4 Assessment of the Level of Potential Negative Impact Based on the Standardized FSS Density Using a 3-Year Moving Average (updated FSEIS Table I-10).
Species Best Fit Slope Slope 1 Slope 2 Final from from from Decision Linear Segmented Segmented Regression Regression Regression Alewife SR S1=0 S2<0 4 American Shad SR S1=0 S2<0 4 Atlantic Tomcod SR S1<0 S2=0 4 Bay Anchovy SR S1=0 S2<0 4 Blueback Herring SR S1<0 S2=0 4 Bluefish SR S1>0 S2<0 4 Hogchoker SR S1<0 S2=0 4 Rainbow Smelt SR S1=0 S2<0 4 Striped Bass LR S<0 4 Weakfish LR S<0 4 White Catfish SR S1=0 S2=0 1 White Perch SR S1<0 S2=0 4 2
Table 5. Competing Models Used To Characterize the Standardized River Segment 4 BSS Population Trends of YOY Fish Density Using a 3-Year Moving Average (updated FSEIS Table I-12).
Species Linear Regression Segmented Regression MSE Slope Std Err p-value MSE Slope 1 Join Slope 2 of Slope Point Estimate Lower Upper Lower Upper 95% CL 95% CL 95% CL 95% CL Alewife 0.725 0.075 0.024 0.004 0.698 -0.088 0.120 2002 -0.007 0.376 American Shad 0.215 -0.121 0.013 0.000 0.234 -0.151 -0.066 2005 -0.404 0.077 Bay Anchovy 0.844 0.059 0.025 0.029 Failed to Converge Blueback Herring 0.726 -0.075 0.024 0.004 0.665 -0.154 0.369 1994 -0.211 -0.043 Bluefish 1.034 0.013 0.028 0.646 0.915 -0.355 0.083 1997 -0.014 0.224 Hogchoker 0.776 0.069 0.024 0.010 0.331 -0.251 -0.023 1998 0.152 0.310 Spottail Shiner 0.993 -0.030 0.028 0.290 0.930 -0.530 2.306 1989 -0.123 0.012 Striped Bass 1.023 0.019 0.028 0.502 0.447 0.110 0.343 1999 -0.292 -0.085 White Perch 1.039 -0.009 0.028 0.753 0.906 -1.099 0.153 1991 -0.042 0.111 Table 6. River Segment 4 Assessment of the Level of Potential Negative Impact Based on the Standardized BSS Density Using a 3-Year Moving Average (updated FSEIS Table I-13).
Species Best Fit Slope Slope 1 Slope 2 Final from from from Decision Linear Segmented Segmented Regression Regression Regression Alewife SR S1=0 S2=0 1 American Shad LR S<0 4 Bay Anchovy LR S>0 1 Blueback Herring SR S1=0 S2<0 4 Bluefish SR S1=0 S2=0 1 Hogchoker SR S1<0 S2>0 4 Spottail Shiner SR S1=0 S2=0 1 Striped Bass SR S1>0 S2<0 4 White Perch SR S1=0 S2=0 1 3
Table 7. Competing Models Used To Characterize the Standardized River Segment 4 LRS Population Trends of YOY Atlantic Tomcod Density Using a 3-Year Moving Average (updated FSEIS Table I-15).
Species Linear Regression Segmented Regression MSE Slope Std Err p-value MSE Slope 1 Join Slope 2 of Slope Point Estimate Lower Upper Lower Upper 95% CL 95% CL 95% CL 95% CL Atlantic Tomcod 1.031 -0.015 0.028 0.604 0.480 -2.709 -0.671 1989 -0.007 0.090 Table 8. River Segment 4 Assessment of the Level of Potential Negative Impact Based on the Standardized LRS Atlantic Tomcod YOY Density Using a 3-Year Moving Average (updated FSEIS Table I-16).
Species Best Fit Slope Slope 1 Slope 2 Final from from from Decision Linear Segmented Segmented Regression Regression Regression Atlantic Tomcod SR S1<0 S2=0 4 4
Table 9. Competing Models Used To Characterize the Standardized River Segment 4, FSS Population Trends of YOY Fish CPUE (updated FSEIS Table I-19).
Species Linear Regression Segmented Regression MSE Slope Std Err p-value MSE Slope 1 Join Slope 2 of Slope Point Estimate Lower Upper Lower Upper 95% CL 95% CL 95% CL 95% CL Alewife 0.953 -0.036 0.024 0.144 0.987 -0.115 0.130 2000 -0.272 0.071 American Shad 0.776 -0.063 0.022 0.007 0.688 -0.123 0.204 1996 -0.224 -0.038 Atlantic Tomcod 0.791 -0.062 0.022 0.010 0.746 -0.492 0.060 1993 -0.090 0.059 Bay Anchovy 1.039 0.003 0.025 0.903 0.882 -0.027 0.231 1999 -0.267 0.021 Blueback Herring 0.938 -0.039 0.024 0.112 0.633 -2.448 -0.120 1987 -0.048 0.049 Bluefish 0.861 -0.052 0.023 0.031 0.847 -0.098 0.090 2002 -0.371 0.049 Hogchoker 0.832 -0.056 0.023 0.019 0.805 -1.988 3.262 1987 -0.125 -0.022 Rainbow Smelt 0.839 -0.055 0.023 0.022 0.837 -0.265 0.451 1992 -0.163 -0.016 Striped Bass 0.952 -0.037 0.024 0.140 0.893 -1.943 3.587 1987 -0.111 -0.003 Weakfish 1.014 -0.020 0.025 0.430 0.967 -0.091 0.130 2000 -0.296 0.092 White Perch 0.948 -0.038 0.024 0.131 0.944 -0.318 0.065 1996 -0.091 0.127 Table10. River Segment 4 Assessment of the Level of Potential Negative Impact Based on the Standardized FSS CPUE (updated FSEIS Table I-20).
Species Best Fit Slope Slope 1 Slope 2 Final from from from Decision Linear Segmented Segmented Regression Regression Regression Alewife LR S=0 1 American Shad SR S1=0 S2<0 4 Atlantic Tomcod SR S1=0 S2=0 1 Bay Anchovy SR S1=0 S2=0 1 Blueback Herring SR S1<0 S2=0 4 Bluefish SR S1=0 S2=0 1 Hogchoker SR S1=0 S2<0 4 Rainbow Smelt SR S1=0 S2<0 4 Striped Bass(a) SR S1=0 S2=0 1 Weakfish SR S1=0 S2=0 1 White Perch SR S1=0 S2=0 1 (a)
Please see the discussion of the regression analyses for striped bass following Table 26 below.
5
Table 11. Competing Models Used To Characterize the Standardized River Segment 4 LRS Population Trends of YOY Atlantic Tomcod CPUE Using a 3-Year Moving Average (updated FSEIS Table I-22).
Species Linear Regression Segmented Regression MSE Slope Std Err p-value MSE Slope 1 Join Slope 2 of Slope Point Estimate Lower Upper Lower Upper 95% CL 95% CL 95% CL 95% CL Atlantic Tomcod 1.012 -0.021 0.025 0.410 0.842 -1.609 0.089 1988 -0.044 0.076 Table 12. River Segment 4 Assessment of the Level of Potential Negative Impact Based 7 on the Standardized LRS Atlantic Tomcod YOY CPUE Using a 3-Year Moving Average (updated FSEIS Table I-23).
Species Best Fit Slope Slope 1 Slope 2 Final from from from Decision Linear Segmented Segmented Regression Regression Regression Atlantic Tomcod SR S1=0 S2=0 1 6
Table 13. Assessment of Population Impacts for IP2 and IP3 River Segment 4 (updated FSEIS Table I-24).
Species Density CPUE River Segment FSS BSS LRS FSS LRS Assessment Alewife 4 1 N/A 1 N/A 2.0 American Shad 4 4 N/A 4 N/A 4.0 Atlantic Menhaden N/A N/A N/A N/A N/A Unknown Atlantic Sturgeon N/A N/A N/A N/A N/A Unknown Atlantic Tomcod 4 N/A 4 1 1 2.5 Bay Anchovy 4 1 N/A 1 N/A 2.0 Blueback Herring 4 4 N/A 4 N/A 4.0 Bluefish 4 1 N/A 1 N/A 2.0 Gizzard Shad N/A N/A N/A N/A N/A Unknown Hogchoker 4 4 N/A 4 N/A 4.0 Rainbow Smelt 4 N/A N/A 4 N/A 4.0 Shortnose Sturgeon N/A N/A N/A N/A N/A Unknown Spottail Shiner N/A 1 N/A N/A N/A 1.0 Striped Bass 4 4 N/A 1 N/A 3.0 Weakfish 4 N/A N/A 1 N/A 2.5 White Catfish 1 N/A N/A N/A N/A 1.0 White Perch 4 1 N/A 1 N/A 2.0 Blue Crab N/A N/A N/A N/A N/A Unknown 7
Table 14. Competing Models Used To Characterize the Standardized Riverwide FSS Population Trends of YOY Fish CPUE (updated FSEIS Table I-27).
Species Linear Regression Segmented Regression MSE Slope Std Err p-value MSE Slope 1 Join Slope 2 of Slope Point Estimate Lower Upper Lower Upper 95% CL 95% CL 95% CL 95% CL Alewife 1.005 0.023 0.025 0.361 Failed to Converge American Shad 0.563 -0.085 0.019 0.000 0.537 -0.362 0.597 1989 -0.152 -0.050 Atlantic Tomcod 0.623 -0.080 0.020 0.000 0.628 . . 1986 -0.128 -0.043 Bay Anchovy 1.030 0.012 0.025 0.633 0.929 -0.012 0.114 2007 -1.362 0.421 Blueback Herring 0.750 -0.066 0.021 0.005 0.795 -0.161 -0.006 2004 -0.279 0.291 Bluefish 0.750 -0.067 0.021 0.005 0.784 -0.125 0.094 1999 -0.268 0.038 Hogchoker 0.941 -0.039 0.024 0.118 0.923 -0.224 0.014 1999 -0.120 0.213 Spottail Shiner 0.838 -0.055 0.023 0.022 0.891 -0.186 0.026 2000 -0.201 0.172 Striped Bass 0.833 -0.056 0.023 0.020 0.590 0.164 2.411 1987 -0.135 -0.042 White Perch 1.022 -0.017 0.025 0.514 1.008 -0.425 0.112 1993 -0.068 0.121 Table 15. Riverwide Assessment of the Level of Potential Negative Impact Based on the Standardized FSS CPUE (updated FSEIS Table I-28).
Species Best Fit Slope Slope 1 Slope 2 Final from from from Decision Linear Segmented Segmented Regression Regression Regression Alewife LR S=0 1 American Shad SR S1=0 S2<0 4 Atlantic Tomcod LR S<0 4 Bay Anchovy SR S1=0 S2=0 1 Blueback Herring LR S<0 4 Bluefish LR S<0 4 Hogchoker SR S1=0 S2=0 1 Spottail Shiner LR S<0 4 Striped Bass SR S1>0 S2<0 4 White Perch SR S1=0 S2=0 1 8
Table 16. Competing Models Used To Characterize the Standardized Riverwide BSS Population Trends of YOY Fish CPUE (updated FSEIS Table I-30).
Species Linear Regression Segmented Regression MSE Slope Std Err p-value MSE Slope 1 Join Slope 2 of Slope Point Estimate Lower Upper Lower Upper 95% CL 95% CL 95% CL 95% CL Alewife 0.701 0.072 0.021 0.002 0.687 -0.054 0.116 2002 -0.012 0.366 American Shad 0.584 -0.083 0.019 0.000 0.562 -0.358 0.623 1990 -0.161 -0.057 Atlantic Tomcod 0.507 -0.090 0.018 0.000 0.390 -0.388 -0.103 1994 -0.092 0.036 Bay Anchovy 0.800 0.061 0.022 0.011 0.593 -0.044 0.063 2006 -0.018 0.989 Blueback Herring 1.040 0.002 0.025 0.946 1.058 -1.084 1.925 1988 -0.080 0.046 Bluefish 1.036 0.008 0.025 0.754 1.048 -0.065 0.188 1999 -0.239 0.115 Hogchoker 1.025 0.015 0.025 0.546 1.054 -0.183 0.098 1998 -0.084 0.231 Rainbow Smelt 0.975 -0.031 0.024 0.210 1.005 -0.270 0.370 1993 -0.138 0.036 Spottail Shiner 0.751 0.066 0.021 0.005 0.811 -0.014 0.155 2003 -0.190 0.291 Striped Bass 1.031 0.012 0.025 0.641 0.930 -0.019 0.125 2005 -0.722 0.232 Weakfish 0.915 -0.044 0.024 0.076 0.480 . . 1986 -0.052 0.023 White Catfish 1.034 -0.010 0.025 0.703 1.008 -1.315 0.543 1989 -0.047 0.084 White Perch 1.030 -0.012 0.025 0.624 1.000 -0.371 0.084 1994 -0.060 0.144 Table 17. Riverwide Assessment of the Level of Potential Negative Impact Based on the BSS CPUE (updated FSEIS Table I-31).
Species Best Fit Slope Slope 1 Slope 2 Final from from from Decision Linear Segmented Segmented Regression Regression Regression Alewife SR S1=0 S2=0 1 American Shad SR S1=0 S2<0 4 Atlantic Tomcod SR S1<0 S2=0 4 Bay Anchovy SR S1=0 S2=0 1 Blueback Herring LR S=0 1 Bluefish LR S=0 1 Hogchoker LR S=0 1 Rainbow Smelt LR S=0 1 Spottail Shiner LR S>0 1 Striped Bass SR S1=0 S2=0 1 Weakfish LR S=0 1 White Catfish SR S1=0 S2=0 1 White Perch SR S1=0 S2=0 1 9
Table 18. Competing Models Used To Characterize the Standardized Riverwide LRS Population Trend of YOY Atlantic Tomcod CPUE (updated FSEIS Table I-33).
Species Linear Regression Segmented Regression MSE Slope Std Err p-value MSE Slope 1 Join Slope 2 of Slope Point Estimate Lower Upper Lower Upper 95% CL 95% CL 95% CL 95% CL Atlantic Tomcod 0.831 -0.057 0.023 0.019 0.880 -0.360 0.141 1994 -0.123 0.053 Table 19. Riverwide Assessment of the Level of Potential Negative Impact Based on the Standardized LRS CPUE of Atlantic Tomcod (updated FSEIS Table I-34).
Species Best Fit Slope Slope 1 Slope 2 Final from from from Decision Linear Segmented Segmented Regression Regression Regression Atlantic Tomcod LR S<0 4 10
Table 20. Competing Models Used To Characterize the Standardized Riverwide YOY Abundance Index Trends (updated FSEIS Table I-36).
Species Linear Regression Segmented Regression MSE Slope Std Err p-value MSE Slope 1 Join Slope 2 of Slope Point Estimate Lower Upper Lower Upper 95% CL 95% CL 95% CL 95% CL Alewife 1.017 0.019 0.025 0.458 0.992 -0.727 0.258 1990 -0.021 0.127 American Shad 0.596 -0.082 0.019 0.000 0.594 -0.379 0.630 1989 -0.153 -0.046 Atlantic Tomcod 0.588 -0.083 0.019 0.000 0.561 -1.665 2.716 1987 -0.141 -0.055 Bay Anchovy 0.744 -0.067 0.021 0.004 0.786 -0.194 0.005 2000 -0.198 0.152 Blueback Herring 0.792 -0.062 0.022 0.010 0.794 -0.154 -0.021 2006 -0.300 0.581 Bluefish 1.035 0.009 0.025 0.731 0.967 -0.038 0.205 1999 -0.259 0.081 Hogchoker 0.902 -0.046 0.023 0.062 0.942 -0.165 0.017 2003 -0.219 0.299 Rainbow Smelt 0.971 -0.033 0.024 0.193 0.960 -0.286 0.480 1992 -0.142 0.015 Spottail Shiner 0.844 0.055 0.023 0.024 0.879 -0.008 0.168 2003 -0.273 0.227 Striped Bass 1.039 0.004 0.025 0.864 0.928 -0.018 0.149 2003 -0.505 0.110 Weakfish 0.647 -0.077 0.020 0.001 0.576 -0.561 0.032 1992 -0.095 0.027 White Catfish 0.833 -0.056 0.023 0.020 0.863 -0.198 0.011 2001 -0.173 0.193 White Perch 1.039 -0.003 0.025 0.906 1.093 -0.079 0.103 2003 -0.424 0.243 Table 21. Riverwide Assessment of the Level of Potential Negative Impact Based in the Abundance Index (updated FSEIS Table I-37).
Species Best Fit Slope Slope 1 Slope 2 Final from from from Decision Linear Segmented Segmented Regression Regression Regression Alewife SR S1=0 S2=0 1 American Shad SR S1=0 S2<0 4 Atlantic Tomcod SR S1=0 S2<0 4 Bay Anchovy LR S<0 4 Blueback Herring LR S<0 4 Bluefish SR S1=0 S2=0 1 Hogchoker LR S=0 1 Rainbow Smelt SR S1=0 S2=0 1 Spottail Shiner LR S>0 1 Striped Bass SR S1=0 S2=0 1 Weakfish SR S1=0 S2=0 1 White Catfish LR S<0 4 White Perch LR S=0 1 11
Table 22. Assessment of Riverwide Population Impacts (updated FSEIS Table I-38).
Species CPUE Abundance Riverwide Index Assessment FSS BSS LRS Alewife 1 1 N/A 1 1.0 American Shad 4 4 N/A 4 4.0 Atlantic Menhaden N/A N/A N/A N/A Unknown Atlantic Sturgeon N/A N/A N/A N/A Unknown Atlantic Tomcod 4 4 4 4 4.0 Bay Anchovy 1 1 N/A 4 2.0 Blueback Herring 4 1 N/A 4 3.0 Bluefish 4 1 N/A 1 2.0 Gizzard Shad N/A N/A N/A N/A Unknown Hogchoker 1 1 N/A 1 1.0 Rainbow Smelt N/A 1 N/A 1 1.0 Shortnose Sturgeon N/A N/A N/A N/A Unknown Spottail Shiner 4 1 N/A 1 2.0 Striped Bass 4 1 N/A 1 2.0 Weakfish N/A 1 N/A 1 1.0 White Catfish N/A 1 N/A 4 2.5 White Perch 1 1 N/A 1 1.0 Blue Crab N/A N/A N/A N/A Unknown 12
Table 23. Weight of Evidence Results for the Population Trend Line of Evidence (updated FSEIS Table H-15).
Species River Riverwide WOE Impact Segment Assessment Score Conclusion Assessment Score Score Alewife 2.0 1.0 1.6 Undetected Decline American Shad 4.0 4.0 4.0 Detected Decline Atlantic Menhaden Unknown Unknown Unknown Unresolved Atlantic Sturgeon Unknown Unknown Unknown Unresolved Atlantic Tomcod 2.5 4.0 3.1 Detected Decline Bay Anchovy 2.0 2.0 2.0 Undetected Decline Blueback Herring 4.0 3.0 3.6 Detected Decline Bluefish 2.0 2.0 2.0 Undetected Decline Gizzard Shad Unknown Unknown Unknown Unresolved Hogchoker 4.0 1.0 2.8 Variable Rainbow Smelt 4.0 1.0 2.8 Variable Shortnose Sturgeon Unknown Unknown Unknown Unresolved Spottail Shiner 1.0 2.0 1.4 Undetected Decline Striped Bass 3.0 2.0 2.6 Variable Weakfish 2.5 1.0 1.9 Undetected Decline White Catfish 1.0 2.5 1.6 Undetected Decline White Perch 2.0 1.0 1.6 Undetected Decline Blue Crab Unknown Unknown Unknown Unresolved 13
Table 24. Parameter Values Used in the Monte Carlo Simulation (updated FSEIS Table I-46).
RIS Survey Linear Slope Error CV of EMR IMR Used Slope plus Mean Density (r) Standard Square Data Error of from (1985-the Slope Regression 1996)
Estimate Alewife BSS 0.075 0.099 0.725 1.294 0.095 0.0020 American Shad BSS -0.121 -0.108 0.215 0.494 0.042 0.0005 Atlantic Tomcod FSS -0.082 -0.059 0.666 0.812 0.036 0.0300 Bay Anchovy FSS -0.088 -0.067 0.601 0.510 0.213 0.0040 Blueback Herring BSS -0.075 -0.051 0.726 1.034 0.095 0.0040 Bluefish BSS 0.013 0.041 1.034 0.754 0.003 0.0005 Hogchoker FSS -0.104 -0.086 0.434 1.224 0.386 0.0005 Rainbow Smelt FSS -0.086 -0.064 0.623 1.211 0.258 0.0005 Spottail Shiner BSS -0.030 -0.002 0.993 1.196 0.031 0.0070 Striped Bass BSS 0.019 0.047 1.023 0.515 0.106 0.0080 Weakfish FSS -0.064 -0.039 0.810 0.637 0.544 0.0005 White Catfish FSS -0.042 -0.016 0.941 2.161 0.114 0.0005 White Perch BSS -0.009 0.019 1.039 1.002 0.076 0.0320 14
Table 25. Quartiles of the Relative Difference in Cumulative Abundance and Conclusions for the Strength-of-Connection From the Monte Carlo Simulation (updated FSEIS Table I-47).
Taxa Number N0 = 1000 N0 = 1 x 108 Strength of of Median Q1 Q3 Median Q1 Q3 Connection Years Conclusion Alewife 20 -0.07 -1.19 1.03 -0.07 -1.17 1.01 Low 27 -0.32 -1.63 1.02 -0.32 -1.69 1.04 American Shad 20 0.06 0.00 0.13 0.06 -0.01 0.13 Low 27 0.05 0.00 0.11 0.06 0.00 0.11 Atlantic Tomcod 20 0.15 -0.03 0.34 0.16 -0.03 0.35 Low 27 0.15 0.01 0.30 0.15 0.01 0.29 Bay Anchovy 20 0.29 0.13 0.44 0.29 0.13 0.44 High 27 0.27 0.15 0.39 0.27 0.15 0.39 Blueback Herring 20 0.21 -0.03 0.46 0.22 -0.02 0.46 Low 27 0.22 0.04 0.41 0.23 0.04 0.42 Bluefish 20 0.45 -0.09 0.99 0.45 -0.09 0.98 Low 27 0.67 0.11 1.21 0.69 0.15 1.23 Hogchoker 20 0.58 0.31 0.85 0.57 0.30 0.86 High 27 0.56 0.35 0.78 0.56 0.35 0.78 Rainbow Smelt 20 0.45 0.16 0.74 0.46 0.16 0.76 High 27 0.45 0.23 0.68 0.45 0.23 0.68 Spottail Shiner 20 0.27 -0.14 0.69 0.28 -0.14 0.70 Low 27 0.34 -0.01 0.70 0.35 0.00 0.69 Striped Bass 20 0.80 0.23 1.36 0.79 0.22 1.36 High 27 1.19 0.60 1.81 1.20 0.60 1.80 Weakfish 20 0.74 0.44 1.06 0.75 0.43 1.06 High 27 0.76 0.50 1.01 0.76 0.51 1.01 White Catfish 20 0.41 -0.18 1.01 0.43 -0.19 1.04 Low 27 0.49 0.00 0.98 0.46 -0.03 0.98 White Perch 20 0.42 -0.09 0.93 0.42 -0.07 0.91 Low 27 0.53 0.10 0.98 0.53 0.08 0.98 15
Table 26. Impingement and Entrainment Impact Summary for Hudson River YOY RIS (updated FSEIS Table H-17).
Species Population Trend Strength of Impacts of IP2 and Line of Evidence Connection IP3 Cooling Systems Line of Evidence on YOY RIS Alewife Undetected Decline Low Small American Shad Detected Decline Low Small Atlantic Menhaden Unresolved Low(b) Small Atlantic Sturgeon Unresolved Low(b) Small Atlantic Tomcod Detected Decline Low Small Bay Anchovy Undetected Decline High Small Blueback Herring Detected Decline Low Small Bluefish Undetected Decline Low Small Gizzard Shad Unresolved Low(b) Small Hogchoker Variable High Moderate Rainbow Smelt Variable High Moderate Shortnose Sturgeon Unresolved Low(b) Small Spottail Shiner Undetected Decline Low Small Striped Bass Variable High Moderate Weakfish Undetected Decline High Small White Catfish Undetected Decline Low Small White Perch Undetected Decline Low Small Blue Crab Unresolved Low(b) Small (b)
Strength of connection could not be established using Monte Carlo Simulation; therefore, strength of connection was based on the rate of entrainment and impingement.
16
Discussion The conclusions presented in the February 2014 report, as well as the preceding tables, were based upon our attempt to recreate and apply the methods used by NRC in developing the conclusions presented in the FSEIS. Applying those methods to the segmented regression analysis of the River Segment 4, FSS CPUE index of abundance for striped bass, the slope 2 estimate from the February 2014 report was considered not statistically significant, but the slope 2 estimate from the current analysis is considered significant. The segmented regression analysis conducted for the February 2014 report produced a 95%
confidence interval for slope 2 from -0.1149 to 0.0005, which was considered not significant because it included zero. Whereas the current segmented regression analysis produced a 95% confidence interval from
-0.1113 to -0.0028, which was considered significant because it did not include zero.
Given the sensitivity that the overall impact conclusions can have to such small changes in statistical results, the statistical methods on which the results were based warranted a careful review. Statistical properties of the FSEIS method for determining which trend slopes are significant are discussed below along with a discussion of the use of the results of the linear regression for the striped bass data.
On page I-9 of Appendix I of the FSEIS, NRC describes its method for determining which trend slopes it considered to be significant, and how it selected a regression model on which to base trend scores:
The statistics displayed in the second table included the mean squared error (MSE) for each model; the estimate of the linear slope and associated 95 percent confidence interval; the p-value associated with the significance test of the null hypothesis that the slope (S) associated with the simple linear model equals zero; the estimated 95 percent confidence interval (CI) of the two slopes from the segmented regression (Slope 1=S1 and Slope 2=S2); and the estimated join point. For the segmented regression, slopes were defined as significant if the CI did not include zero.
The best-fit model (defined as the model with the smaller MSE) was then characterized in a third table, based on the general trend depicted by the direction of the estimated slopes. If the slope was significantly different from 0, the trend was represented by either the statement S > 0 for a positive slope or S < 0 for a negative slope. If the slope was not significant, the statement depicting the lack of a trend was S = 0.
For the segmented regressions, the FSEIS test of the null hypothesis that a slope is equal to zero (based on the confidence interval including zero) is based on the standard error of the slope estimate and is equivalent to a t-test of the significance of the slope estimate. That test is a conditional test addressing the question: after all other parameters in the model are estimated, does the inclusion of the slope estimate in the model significantly improve the fit of the model to the data? The result of the test depends on the estimates of all of the other parameters as well as the estimate of the slope being tested. Accordingly, the reliability of the other parameter estimates may affect the result of the significance test for the slope. For the striped bass, Region 4, FSS, CPUE example, NRCs method could lead to the conclusion that slope 2 was significant, although none of the other parameter estimates would have been considered statistically significant:
17
Parameter Lower 95% Upper 95%
Confidence Limit Confidence Limit Join Point 1982 1992 Slope 1 -1.9426 3.5866 Slope 2 -0.1113 -0.0028 In addition, the FSEIS test of significant slopes does not necessarily provide an error rate of 5%, as suggested by a test based on 95% confidence intervals. As noted above, the test is equivalent to a t-test of the significance of the slope estimate. For each test based on the 95% confidence interval, the probability of falsely rejecting the null hypothesis when it is true is 5%. That is the case if one test is performed.
However, if two tests are performed (e.g. the test for slope 1 and the test for slope 2) then the probability of falsely rejecting the null hypothesis in at least one test, when both slopes in fact are zero, is greater than 0.05. For example, the probability of correctly failing to reject the null hypothesis in two independent tests when the null hypothesis is true (e.g., when both slopes are zero) is 0.95 x 0.95. In this case the probability of falsely rejecting the null hypothesis is 1 - (0.95 x 0.95) = 0.0975, almost twice the intended error probability of 0.05.
For regressions with multiple parameters being estimated simultaneously, the two issues discussed above can be avoided by testing the significance of all parameter estimates simultaneously. By doing so, the reliability of every parameter estimate is taken into account. In addition, the probability of falsely rejecting the null hypothesis is controlled and not affected by the number of parameters in the model. The standard F-test for the overall significance of a regression model can be used for this purpose. For this test, the null hypothesis is that all slopes in the model are equal to zero. The alternative hypothesis is that at least one parameter is not equal to zero (Draper, N.R. and H. Smith, 1966. Applied Regression Analysis.). If the regression is not statistically significant based on the standard F-test, then the null hypothesis that all parameters are zero cannot be rejected.
For these reasons, the standard F-test for a regression model is the appropriate test for the null hypothesis that both slopes in the segmented regression are equal to zero. The segmented regression has three parameters in addition to the intercept: two slopes and the join point. By labeling the first year in the time series as year zero, the join point parameter is equal to zero when there is no change in slope over the years being analyzed. Therefore the null hypothesis of all parameters being equal to zero would be: slope 1
= 0, slope 2 = 0 and join point = 0 (i.e., no change in slope). If the null hypothesis is rejected by the F-test, then the FSEIS conditional method could be applied to identify which particular slope estimates are significantly non-zero.
For the striped bass, River Segment 4, FSS, CPUE example, the F-statistic for the segmented regression is 2.04 (with 3 and 23 degrees of freedom). The probability of observing an F-statistic that large under the null hypothesis is 0.14. Because 0.14 exceeds the acceptable error rate for the test of 0.05, the null hypothesis that both slopes are equal to zero cannot be rejected. Because the null hypothesis that all slopes are equal to zero cannot be rejected, it is not necessary or appropriate to apply the FSEIS conditional test for the significance of individual slopes in this case. This more robust statistical result that neither slope 1 nor slope 2 is significantly different from zero is consistent with the results presented in the February 2014 report.
18